/*
Michele Margolis	
Proposal title: Evangelical or Born-Again Christian: Unpacking a Double-Barreled Question


HYPOTHESES

Stated-Hyp1: The demographics of those who identify as Evangelical Christian based on three different question/response specifications will vary.
"I will look at basic demographic, religious, and political information—such as, gender, race, education, income, region of residence, church attendance, self-reported religiosity and spirituality, party identification, and political ideology—to see how the demographic makeup of the evangelical group differs as a function of question wording" (p. 3)
	 

	Test-Hyp1: The share of self-reported Evangelicals will vary between double-barelled vs. separate questions, born-again first.	
	
	Test-Hyp2: The share of women will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	

	Test-Hyp3: The level of education will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	

	Test-Hyp4: The level of income will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	
	Test-Hyp5: The level of republican identification will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	

	Test-Hyp6: The level of conservative identification will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
		
	Test-Hyp7: The share of self-reported Evangelicals will vary between double-barelled vs. separate questions, evangelical first.	
	
	Test-Hyp8: The share of women will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	

	Test-Hyp9: The level of education will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.			

	Test-Hyp10: The level of income will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	
	
	Test-Hyp11: The level of republican identification will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	

	Test-Hyp12: The level of conservative identification will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.		
	
	Test-Hyp13: The share of self-reported Evangelicals will vary between born again vs. evangelical first (separate questions).	
	
	Test-Hyp14: The share of women will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	

	Test-Hyp15: The level of education will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	

	Test-Hyp16: The level of income will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	
	Test-Hyp17: The level of republican identification will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	

	Test-Hyp18: The level of conservative identification will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).			

	Test-Hyp19: The level of religiosity will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.		

	Test-Hyp20: The level of religiosity will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.			

	Test-Hyp21: The level of religiosity will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).		
	
	Test-Hyp22: The level of spirituality will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.		

	Test-Hyp23: The level of spirituality will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.			

	Test-Hyp24: The level of spirituality will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).		
		
	Test-Hyp25: Share of whites vs. blacks will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.		

	Test-Hyp26: Share of whites vs. blacks will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.			

	Test-Hyp27: Share of whites vs. blacks will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).			
	
	

Stated-Hyp2: The attitudes of those who identify as Evangelical Christian based on three different question/response specifications will vary. (p. 3)

	Test-Hyp28: Views on abortion will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	
	Test-Hyp29: Views on immigration will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	

	Test-Hyp30: Racial attitudes will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	

		
	Test-Hyp31: Views on abortion will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	
	
	Test-Hyp32: Views on immigration will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	

	Test-Hyp33: Racial attitudes will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.		
		
	Test-Hyp34: Views on abortion will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	
	Test-Hyp35: Views on immigration will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	

	Test-Hyp36: Racial attitudes will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	

	
Stated-Hyp2: TESS public affiars profile responses of those who identify as Evangelical Christian based on three different question/response specifications will vary. (p. 3)
	
	Test-Hyp37: Interest in politics will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	
	Test-Hyp38: Interest in politics will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	

	Test-Hyp39: Interest in politics will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	

********************************************************************************
NOTES
* bonferonni correction needed for hypotheses 2-6, 8-12, and so on

Discussed in meeting on 3/8/2021.

How to analyze (from proposal):
"I will look at basic demographic, religious, and political information—such as, 
gender, race, education, income, region of residence, church attendance, 
self-reported religiosity and spirituality, party identification, and political 
ideology—to see how the demographic makeup of the evangelical group differs as 
a function of question wording" (p. 3) 

- We decided to do the straight-forward hypotheses. Recode education as a 
continuous variable. (see protocol). Treat race as black vs. white (see protocol).
- Outcome is yes/no evangelical; and all the demographics are moderators.

- Recode abortion such that mixed response is in the middle.

How to analyze (from proposal):
"I will look at additional questions including questions from TESS public 
affairs profile." 

The author requested the following public affiars questions (apart from 
religiosity and spirituality covered previously)
PA006 - politics attention
PA015ANEW - how voted
PA015NEW - how voted
PAC1 - where born

- Decided to not pursue where born and voting because of categorical 
responses.
*/

clear all
use "Margolis1204.dta", clear

********************************************************************************

* INDICATORS OF EXPERIMENTAL MANIPULATIONS
	* see quex for details

	tab Q_ASKED
	lab def treat 1 "1. double barelled" 2 "2. born again first" 3 "3. evangelical first"
	gen treatment = Q_ASKED
	lab val treatment treat
	tab treatment
	
	* did R identify as evangelical (or born-again in double-barelled)
	gen evangelical = 1 if Q1==1
	replace evangelical=0 if Q1==2
	replace evangelical=1 if Q3==1
	replace evangelical=0 if Q3==2
	tab evangelical
	
* OUTCOME MEASURES

* respondent gender
	tab GENDER
	recode GENDER (1=1) (2=0), gen(male)

* respondent education
	tab EDUC // note: education is a categorical variable but we're treating it as continuous. 
	recode EDUC (1/4=8) (5=9) (7/8=11) (9=12) (10/11=14) (12=16) (13=18) (14=20)
	
* respondent income 
	tab INCOME // note: income is a categorical variable but we're treating it as continuous. 

* respondent political party identification
	tab P_PARTYID

* respondent political ideology
	tab P_IDEO
	// recoding last category to missing
	replace P_IDEO = . if P_IDEO>7
	tab P_IDEO
 
* race (see protocol for guidance on coding) 
	recode RACETHNICITY (1=1) (2=0) (*=.), gen(whitevsblack)
	tab whitevsblack
 
* religiosity
	tab PA025
	clonevar religiosity = PA025
	replace religiosity=. if PA025>4
	tab religiosity

* spirituality 
	tab PA026
	clonevar spirituality = PA026
	replace spirituality=. if PA026>4
	tab spirituality 
 
* abortion views
	tab Q8
	recode Q8 (1=1) (3=2) (2=3) (*=.), gen(prochoice)
	tab prochoice

* immigration views
	foreach var in Q7A Q7B Q7C Q7D {
		replace `var'=. if `var'>5
	tab `var'
	}
	* recoding so higher values indicate support
	gen revQ7C= 6-Q7C
	
	gen immigration_support = Q7A + Q7B + revQ7C + Q7D
	tab immigration_support

* racial attitudes
	/*Please indicate to what extent you agree or disagree with the following statement: Discrimination against whites has become as big a problem as discrimination against blacks and other minorities.
	*/

	tab Q4
	replace Q4=. if Q4>5
	clonevar antiracist=Q4

* public affiars profile questions requested:

*PA006 - politics interest
tab PA006

********************************************************************************

* ANALYSIS
	

*Test-Hyp1: The share of self-reported Evangelicals will vary between double-barelled vs. separate questions, born-again first.	
	reg evangelical i.treatment if treatment!=3
	// do not reject. 0.000
	tess 2.treatment, init(Margolis1204)
	
*Test-Hyp2: The share of women will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	reg  male i.treatment if treatment!=3 & evangelical==1
	// reject. 0.449
	tess 2.treatment, bonf(12)

*Test-Hyp3: The level of education will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	reg  EDUC i.treatment if treatment!=3 & evangelical==1
	// reject. 0.010
	tess 2.treatment, bonf(12)
	
*Test-Hyp4: The level of income will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	reg  INCOME i.treatment if treatment!=3 & evangelical==1
	// reject. 0.026 
	tess 2.treatment, bonf(12)
	
*Test-Hyp5: The level of republican identification will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	reg  P_PARTYID i.treatment if treatment!=3 & evangelical==1
	// reject. 0.049
	tess 2.treatment, bonf(12)

*Test-Hyp6: The level of conservative identification will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	reg  P_IDEO i.treatment if treatment!=3 & evangelical==1
	// reject. 0.138
	tess 2.treatment, bonf(12)	
	
	
		
*Test-Hyp7: The share of self-reported Evangelicals will vary between double-barelled vs. separate questions, evangelical first.
	reg evangelical i.treatment if treatment!=2
	// do not reject. 0.000
	tess 3.treatment
	
	
	
*Test-Hyp8: The share of women will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	
	reg  male i.treatment if treatment!=2 & evangelical==1
	// reject. 0.086
	tess 3.treatment, bonf(12)

*Test-Hyp9: The level of education will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.			
	reg  EDUC i.treatment if treatment!=2 & evangelical==1
	// reject. 0.033
	tess 3.treatment, bonf(12)

*Test-Hyp10: The level of income will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	
	reg  INCOME i.treatment if treatment!=2 & evangelical==1
	// reject. 0.069
	tess 3.treatment, bonf(12) 

*Test-Hyp11: The level of republican identification will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	
	reg  P_PARTYID i.treatment if treatment!=2 & evangelical==1
	// reject. 0.042
	tess 3.treatment, bonf(12)
	
*Test-Hyp12: The level of conservative identification will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.		
	reg  P_IDEO i.treatment if treatment!=2 & evangelical==1
	// reject. 0.027
	tess 3.treatment, bonf(12) 

	
	
*Test-Hyp13: The share of self-reported Evangelicals will vary between born again vs. evangelical first (separate questions).	
	reg evangelical i.treatment if treatment!=1
	// do not reject. 0.124
	tess 3.treatment
	
	
	
*Test-Hyp14: The share of women will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg  male i.treatment if treatment!=1 & evangelical==1
	// reject. 0.398
	tess 3.treatment, bonf(12)
	
*Test-Hyp15: The level of education will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg  EDUC i.treatment if treatment!=1 & evangelical==1
	// reject. 0.617
	tess 3.treatment, bonf(12)
	
*Test-Hyp16: The level of income will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg  INCOME i.treatment if treatment!=1 & evangelical==1
	// reject. 0.648
	tess 3.treatment, bonf(12)
	
*Test-Hyp17: The level of republican identification will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg  P_PARTYID i.treatment if treatment!=1 & evangelical==1
	// reject. 0.991
	tess 3.treatment, bonf(12)
	
*Test-Hyp18: The level of conservative identification will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).			
	reg  P_IDEO i.treatment if treatment!=1 & evangelical==1
	// reject. 0.569
	tess 3.treatment, bonf(12)	

*Test-Hyp19: The level of religiosity will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.		

	reg  religiosity i.treatment if treatment!=3 & evangelical==1
	// reject. .259
	tess 2.treatment, bonf(12)
	
*Test-Hyp20: The level of religiosity will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.			

	reg  religiosity i.treatment if treatment!=2 & evangelical==1
	// reject. .356
	tess 3.treatment, bonf(12)
	
*Test-Hyp21: The level of religiosity will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).		
	
	reg  religiosity i.treatment if treatment!=1 & evangelical==1
	// reject. 	.821
	tess 3.treatment, bonf(12)	
*Test-Hyp22: The level of spirituality will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.		

	reg  spirituality i.treatment if treatment!=3 & evangelical==1
	// reject. .812
	tess 2.treatment, bonf(12)
	
*Test-Hyp23: The level of spirituality will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.			

	reg  spirituality i.treatment if treatment!=2 & evangelical==1
	// reject. .987
	tess 3.treatment, bonf(12)
	
*Test-Hyp24: The level of spirituality will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).		
		
	reg  spirituality i.treatment if treatment!=1 & evangelical==1
	// reject. 	.815		
	tess 3.treatment, bonf(12)
	
*Test-Hyp25: Share of whites vs. blacks will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.		

	reg  whitevsblack i.treatment if treatment!=3 & evangelical==1
	// reject. .253
	tess 2.treatment, bonf(12)
	
*Test-Hyp26: Share of whites vs. blacks will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.			
	reg  whitevsblack i.treatment if treatment!=2 & evangelical==1
	// reject. .133
	tess 3.treatment, bonf(12)
	
*Test-Hyp27: Share of whites vs. blacks will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).		
	reg  whitevsblack i.treatment if treatment!=1 & evangelical==1
	// reject. 	.779	
	tess 3.treatment, bonf(12)	
	
	
*Test-Hyp28: Views on abortion will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	reg prochoice i.treatment if treatment!=3 & evangelical==1
	// do not reject. p=0.002
	tess 2.treatment, bonf(12)	
	
*Test-Hyp29: Views on immigration will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	reg immigration_support i.treatment if treatment!=3 & evangelical==1
	// reject. p=0.726
	tess 2.treatment, bonf(12)	

*Test-Hyp30: Racial attitudes will vary between those who identify as evangelical in double-barelled vs. separate questions, born-again first.	
	reg antiracist i.treatment if treatment!=3 & evangelical==1
	// reject. p=0.643
	tess 2.treatment, bonf(12)
	
*Test-Hyp31: Views on abortion will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	
	reg prochoice i.treatment if treatment!=2 & evangelical==1
	// do not reject. p=0.004
	tess 3.treatment, bonf(12)
	
*Test-Hyp32: Views on immigration will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.	
	reg immigration_support i.treatment if treatment!=2 & evangelical==1
	// reject. p=0.120
	tess 3.treatment, bonf(12)
	
*Test-Hyp33: Racial attitudes will vary between those who identify as evangelical in double-barelled vs. separate questions, evangelical first.		
	reg antiracist i.treatment if treatment!=2 & evangelical==1
	// reject. p=0.697
	tess 3.treatment, bonf(12)
	
*Test-Hyp34: Views on abortion will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg prochoice i.treatment if treatment!=1 & evangelical==1
	// reject. p=0.761
	tess 3.treatment, bonf(12)
	
*Test-Hyp35: Views on immigration will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg immigration_support i.treatment if treatment!=1 & evangelical==1
	// reject. p=0.299
	tess 3.treatment, bonf(12)	
*Test-Hyp36: Racial attitudes will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg antiracist i.treatment if treatment!=1 & evangelical==1
	// reject. p=0.928

	tess 3.treatment, bonf(12)	
	
	
*Test-Hyp37: Interest in politics will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg PA006 i.treatment if treatment!=3 & evangelical==1
	// reject. .283
	tess 2.treatment, bonf(12)
	
*Test-Hyp38: Interest in politics will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg PA006 i.treatment if treatment!=2 & evangelical==1
	// reject. p=0.009 (bonferonni correction - see notes)
	tess 3.treatment, bonf(12)
	
*Test-Hyp39: Interest in politics will vary between those who identify as evangelical in born again vs. evangelical first (separate questions).	
	reg PA006 i.treatment if treatment!=1 & evangelical==1
	// reject.	.173	
	tess 3.treatment, bonf(12)
